Incompatible NumPy Version Declared in pyproject.toml

In pyproject.toml

dependencies = [
"timm>=1.0.17",
"numpy==1.26",
"tqdm",
"ftfy==6.1.1",
"regex",
"iopath>=0.1.10",
"typing_extensions",
"huggingface_hub",
]

The current pyproject.toml in the SAM3 repository pins NumPy to:

numpy == 1.26

This is causing compatibility issues in environments where SAM3 is used — especially on Kaggle and Google Colab because the pinned version is either:

too old for some dependencies
or gets force-upgraded automatically, creating conflicts
or prevents installation in environments that already ship NumPy 2.x

Geeting Error:
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
sam3 0.1.0 requires numpy==1.26, but you have numpy 2.3.5 which is incompatible.
numba 0.60.0 requires numpy<2.1,>=1.22, but you have numpy 2.3.5 which is incompatible.
opencv-python-headless 4.12.0.88 requires numpy<2.3.0,>=2; python_version >= "3.9", but you have numpy 2.3.5 which is incompatible.
tensorflow 2.19.0 requires numpy<2.2.0,>=1.26.0, but you have numpy 2.3.5 which is incompatible.
opencv-python 4.12.0.88 requires numpy<2.3.0,>=2; python_version >= "3.9", but you have numpy 2.3.5 which is incompatible.
opencv-contrib-python 4.12.0.88 requires numpy<2.3.0,>=2; python_version >= "3.9", but you have numpy 2.3.5 which is incompatible.

Start with the dependency declaration in pyproject.toml and reproduce installation in an environment with NumPy 2.x, such as Kaggle or Google Colab. Review the listed dependency conflicts and determine a supported NumPy constraint that does not break the project’s installation; done means the package installs without the reported resolver conflict.

Written by the indexing model from the issue text.